The Role of Management and Safety Climate in Preventing Risk Taking at Work

The role of management and safety climatein preventing risk-taking at work

Steven Yule* and Rhona FlinIndustrial Psychology Research Centre, School of Psychology,University of Aberdeen, Aberdeen AB24 2UB, UKFax: �44 1224 273211 E-mail: [email protected]: [email protected]*Corresponding author

Andy MurdyE.ON UK plc, Westwood Way, Westwood Business Park, Coventry CV4 8LG, UKE-mail: [email protected]

Abstract: Safety climate is a leading performance indicator that can provideinsight into safety performance before accidents have occurred. Managerialvariables have emerged as a primary determinant of safety climate in empiricalresearch. In order to investigate the mechanisms of that influence a theoreticalmodel was developed to test the relationship between management commitmentand worker risk-taking. Workforce perceptions of safety climate (n�1026) werecollected using the Health and Safety Executive climate survey tool andanalysed using Structural Equation Modelling (SEM). The tested modelrevealed that the relationship between management commitment and supervisorinvolvement with risk-taking behaviours was mediated by knowledge andtraining. Additionally, a positive attitude towards risk taking (i.e. not engagingin risk taking behaviours) was related to enhanced feelings of workers’responsibility for safety and more positive appraisals of senior managementcommitment. Managers may find the model useful when attempting to improvesafety climate.

Keywords: continual improvement; risk; safety climate; senior managementcommitment; supervisor involvement.

Reference to this paper should be made as follows: Yule, S., Flin, R. andMurdy, A. (2007) ‘The role of management and safety climate in preventingrisk-taking at work’, Int. J. Risk Assessment and Management, Vol. 7, No. 2,pp.137–151.

Biographical notes: Steven Yule (MA, MSc, PhD) is a Research Fellow in theSchool of Psychology at the University of Aberdeen. His research is on theimpact that non-technical skills (i.e. situation awareness, decision making,teamwork, communication, leadership) can have on the safety performance oforganisations that are exposed to major hazards. He also has a particular interestin safety climate, senior managers’ leadership and patient safety. His currentwork is developing a behavioural marker system with the Royal College ofSurgeons of Edinburgh and the NHS, which will be used to train and ratesurgeons’ non-technical skills in the operating theatre.

Copyright © 2007 Inderscience Enterprises Ltd.

Int. J. Risk Assessment and Management, Vol. 7, No. 2, 2007 137

Rhona Flin (BSc, PhD Psychology) is Professor of Applied Psychology in theSchool of Psychology at the University of Aberdeen. She is a CharteredPsychologist, a Fellow of the British Psychological Society and a Fellow of theRoyal Society of Edinburgh. She directs a team of psychologists working withhigh reliability industries on the management of safety and emergency response.The group has worked on projects relating to aviation safety, leadership andsafety in the offshore industry, team skills and emergency management. She iscurrently studying anaesthetists’ and surgeons’ non-technical skills and thesafety climate in hospitals.

Andy Murdy is Senior Health and Safety Advisor for E.ON UK. He has workedfor E.ON UK and its predecessors since 1987, with a full-time health and safetyrole since 1996. Andy’s career in health and safety has involved work in sixcountries, including a two year post in the USA looking at operations andconstruction health and safety issues. In his current role, his key responsibilitiesinclude performance management, audit and behavioural safety. Andy is acorporate member of IOSH.

1 Introduction

The safety performance of industrial organisations has traditionally been assessed againstlagging criteria such as accident and injury rates. More recently, leading performanceindicators have allowed organisations to assess their safety performance without the needfor retrospective analysis of accidents. Examples of leading performance measuresinclude safety audits, hazard analysis and safety climate. The concept of safety climate isof particular interest to managers, engineers and psychologists as it has a number ofimplications for the relationship between individual and group perceptions of safetymanagement, and behaviours at work which can have an impact on safety performance.A number of studies have found workforce perceptions of safety climate to be directlyand indirectly linked to safety outcomes (Donald and Canter, 1994; Hofmann and Stetzer,1996; Lee, 1998; Mearns et al., 2001; Niskanen, 1994; Tomas et al., 1999; Zohar, 2000).Among these are studies that use Structural Equation Modelling (SEM), a method ofmultivariate analysis in which interrelations between concepts made along theoreticaldimensions can be tested statistically (Byrne, 1994). Authors often interpret the results ofSEM as evidence for causal paths between latent and observed variables.

2 Safety climate

It is generally accepted that safety climate is a ‘snapshot’ of workforce perceptionsabout safety (Mearns et al., 1997). However, researchers are in less agreement regardingwhich safety climate factors or dimensions are most important in influencing behavioursat work. The multiple definitions of safety climate in the literature (Flin et al., 2000;Guldenmund, 2000) have determined to a large extent what variables research teams haveincorporated when developing measures of safety climate. The central debate amongtheorists appears to be whether the safety climate should be restricted to workforceperceptions about management and the manner in which management reconciles safetywith productivity (Brown and Holmes, 1986; Dedobbeleer and Beland, 1991; Zohar,

138 S. Yule, R. Flin and A. Murdy

1980, 2000), or whether the role of management is incorporated with other safety issuessuch as risk perception, worker involvement, personal accountability, perceptions of thephysical environment and job communication (Cheyne et al., 1998; Cox and Cox, 1991;Cox and Flin, 1998; Lee, 1998; Mearns et al., 1998; Mearns et al., 2001; Williamson et al., 1997). This debate has not been resolved, and has led to a proliferation ofinstruments that, with one or two exceptions, are specific to an organisation, or at best, tothe industrial sector for which they were developed. Two notable exceptions include theOffshore Safety Questionnaire (OSQ), developed by Flin et al. (1996) on the basis ofearlier work by Marek et al. (1985), and Rundmo (1994). The OSQ has been used inseveral offshore environments, as well as in healthcare, mining and forestry settingsaround the world. The second exception is the HSE Health and Safety Climate survey tool(Health and Safety Executive, 1997), which was developed by the UK regulator to be used across all UK industrial sectors.

3 Management commitment to safety

Despite the disagreement outlined above, management clearly has a role to play in safetyclimate. A review and thematic analysis of safety climate factors by Flin et al. (2000)found that management was central to 72% of the studies. Management was alsoidentified as one of only two factors (the other being workforce involvement) that wereproperly replicated across studies (Dedobbeleer and Beland, 1998). A third review ofsafety climate themes included a number of studies not covered in the two reviews above, and found that management was the most frequently measured dimension(Guldenmund, 2000).

Management commitment to safety had actually been studied before the term ‘safetyclimate’ was coined. Smith et al. (1978) found that workforce perceptions of a high levelof management commitment to safety were associated with low accident rates in a cross-section of 42 US industrial plants. The seminal study on safety climate by Zohar(1980) identified two dimensions as being most influential in determining safety climatelevel. The second of these (after relevance of safety to job behaviour) was workforceperceptions of management attitudes to safety. Zohar argued that the managementcommitment was a prerequisite of successful initiatives aimed at improving the state ofsafety in industrial organisations. This argument has found considerable empirical supportacross industrial sectors (although mainly in the energy sector and manufacturing). Froma study of workers in UK chemical plants, Donald and Canter (1994) found that a number of safety climate scales correlated with self-reported accident involvement,including management commitment.

Further evidence is provided by Diaz and Cabrera (1997), who reported that safetyclimate differentiated organisations with differing levels of safety in a sample of Spanishairport workers. Workforce perceptions of company safety policy (including managementcommitment) were deemed to be the most important factor, with perceptions of theorganisational philosophy regarding the relative priorities of productivity versus safety.Rundmo (1994) also found that management commitment to safety was the mostimportant determinant of workforce satisfaction with safety, and with safety-relatedcontingency measures. Organisational support provided by management was the secondmost important determinant. Finally, Cheyne et al. (1999) found that management

The role of management and safety climate in preventing risk-taking at work 139

influenced workforce appraisals of commitment across three UK industries:manufacturing, dairy produce and transport. Management also consistently influencedtraining across all three samples but only influenced workforce personal actions andresponsibility in the manufacturing and dairy produce samples.

The applied use of such associations is underlined by a study in Australian healthcare.Coyle et al. (1995) argue that modifying the attitudes of management and workforcetoward health and safety should improve an organisation’s safety climate and ultimatelytheir safety record. These findings are not universal, however.

4 Supervisor involvement in safety

Empirical research also indicates that supervisors have an important role to play in safety climate. For example, a model that integrates the safety influences of managers andsupervisors is offered by Thompson et al. (1998), who tested a model based around twocentral pathways

� from ‘organisational politics’ to ‘manager support for safety’ to ‘safety conditions’

� from ‘supervisor fairness’ to ‘supervisor support for safety’ to ‘safety compliance’.

Management support for safety was also found to positively influence supervisor supportfor safety. They concluded that management has an influence on safety conditions butworkforce compliance with safety rules and regulations under those conditions isinfluenced by the perceived fairness of the supervisor. O’Dea (2002) also found thatsupervisor commitment to safety was predictive of worker propensity to take safetyinitiatives, and comply with rules.

Supervisors have been shown to have other important influences regarding safetyclimate. From three Spanish samples of ‘high risk organisations’, Tomas et al. (1999)found that supervisors played an important role in the accident prevention process bytransferring the elements of safety climate to members of the workforce. Evidence for thiscame from support for a tested model in which the causal chain ran from ‘safety climate’to ‘supervisor response’ to ‘co-worker response’ to ‘worker attitude’, and then to ‘safetybehaviour’, ‘risk’ and finally ‘accidents’. Brown et al. (2000) report the corollary thatsupervisors can have a negative impact on safety climate by applying too much pressureon workers, a conclusion based on a study in the US steel industry.

Zohar (2002) studied some of these concepts experimentally at an Israeli maintenanceplant. He measured the number of safety-related interactions (episodes) betweensupervisors and workers, and gave weekly feedback to supervisors on their performance.As a result, the frequency of safety-related interactions increased rapidly from 9 to 58%of all interactions. In experimental groups there was an associated significant decrease inaccidents, an increase in personal protective equipment (PPE) use (earplugs), and asignificant improvement in safety climate perceptions compared with no change in control groups. This study showed that supervisors could dramatically improve safetyperformance and PPE use by merely emphasising safety in interactions that take place on the shop floor as a matter of course, and is an example of a micro-level change inculture.


5 The role of risk in safety climate

Early research on safety climate incorporated risk as a key factor. For example, Zohar(1980) measured risk in his seminal work, and Brown and Holmes (1986) found that athree factor model of safety climate comprising risk, management concern andmanagement action, had an acceptable fit to workforce safety climate in a US productionsample. A number of studies have used risk an outcome variable. For example,Dedobbeleer and Beland (1991) found that workforce attitudes towards safety practices(i.e. propensity to take risk) were predicted by perceptions of management concern,replicating the Brown and Holmes (1986) findings. Additionally, Tomas et al. (1999)found that perception of actual risk was the only factor to directly predict accidents intheir model outlined earlier. Although there is little research linking worker attitudes tosenior management and risk-taking, Kivimaki et al. (1995) found that workforce trust insenior managers was linked to workers’ risk perception and goal acceptance. More recentresearch on the role of senior managers confirms the conclusion that the attitudes of seniormanagers influence workers’ behavioural intentions, and are related to the achievement ofsafe working practices (Rundmo and Hale, in press). Yule (2003) also found that seniormanagers in the UK and US energy sectors who received higher performance appraisalsfrom their employees were more likely to discuss and understand the risks that theirworkers were exposed to.

The aim of the present paper is to examine the role of managers and supervisors ininfluencing risk-taking behaviours. In order to achieve this goal, a theoretical model ofsafety climate was developed, based on the previous research findings outlined above,along with consideration of the underlying theory. For these reasons, and the fact thatworkers have been found to perceive risks in industrial contexts (Rundmo and Sjöberg,1998) in a valid and reliable manner, perceptions of risk-taking behaviour will be used asan outcome measure in the model.

6 Developing the model

The present section outlines the rationale for developing a theoretical model of safetyclimate. The model is presented in Figure 1 below. Management has been shown to holda significant influence on workforce safety climate (Flin et al., 2000; Kozlowski andDoherty, 1989). Andriessen (1978) also argues that workforce perception of the safetyattitudes of senior management is an important factor in influencing motivation to behavesafely and Clarke (1999) argues that different levels of management have distinct rolesand are perceived differently by the workforce. This distinction is reflected in the currentmodel as two layers of management – senior management and supervisors – are identifiedon the basis of exploratory factor analysis of the workforce safety climate data set. Therelationship between these managerial factors and other factors will form the basis of thetheoretical structural component.

It is plausible that relationships exist between attitudes towards different levels ofmanagement. Both Dedobbeleer and Beland (1991) and Thompson et al. (1998) providetested models that show that attitudes toward first-line supervisors are predicted byattitudes toward senior management. Witt et al. (1994) take the opposite view and testeda model which displays a path running from ‘supervisor fairness’ to ‘management support


for safety’. On the basis of the research evidence, the hypothesised model in the presentpaper argues for a reciprocal relationship between senior managers (F1) and supervisors(F5). This reciprocal relationship is imposed because supervisors implement the majorityof resources, incentives and procedures laid down by senior management for workforcehealth and safety.

In the model, senior managers are thought to have their primary influence on elementsof the safety system (F7), i.e. by establishing and endorsing the Permit-To-Work (PTW)system1, and by providing safety training (F3). Senior management sets the framework ofthe safety system and it is hypothesised that perceptions of their commitment to workforcesafety predicts the perceived efficacy of the procedures in place and the PTW system.Further evidence for this link is provided by the International Nuclear Safety AdvisoryGroup (INSAG). They published a three-level framework for the measurement of safety culture (INSAG, 1991). One of the levels, ‘Managers’ Commitment’ isdistinguished by (among others) ‘definition of responsibilities’; definition and control ofsafety practices’ and ‘qualifications and training’.

The role of knowledge and training in health and safety (F3) is pivotal in the model asit is hypothesised to be directly related to risk taking behaviours and thus to mediate therelationship between senior management commitment and risk taking behaviour. This isproposed to be an inverse relationship – as knowledge increases, the propensity to engagein risk-taking behaviours will decrease. A direct relationship between senior managementcommitment and risk taking behaviour was also specified in the model so that acomparison between that relationship and the mediated relationship could be made.

The model acknowledges previous research that has emphasised the importance ofsupervisor involvement in safety climate. For example, Fleming et al. (1996) suggest thatoffshore supervisors should adopt a participative style of leadership and be involved inworkforce training. Training is one contribution to knowledge so a theoretical path fromsupervisor involvement (F5) to knowledge (F3) was imposed. Simard and Marchand(1994) argue that first-line supervisors who adopt a participative style are also mosteffective in occupational safety and effective at encouraging team members to participatein safety initiatives. Participating in safety initiatives is one way of boosting responsibilityfor safety. Cheyne et al. (1998) also show the influence of management attitudes onworker responsibility, through mediating factors of involvement, communication andhazards in the manufacturing sector. This is in line with the management literature where delegation is an important managerial practice (Yukl, 1998). By delegating,supervisors encourage workers to accept responsibility for safety in the workplace. Thisprinciple provides more evidence for a relationship between supervisor involvement andresponsibility which is stipulated in the model. However, the path was imposed with theawareness that feelings of responsibility may arise also through other sources such as jobcharacteristics (Hackman and Oldham, 1980).

Simard and Marchand (1995) place supervisory characteristics, such as theirexperience and approach to safety management, along with workgroup cohesiveness atthe micro-level of their organisational model. This cooperation between workforce andsupervisor should make it easier for the workforce to make suggestions and exert pressurefor change; see also Fleming et al. (1996). Therefore, a structural path from supervisorinvolvement (F5) to teamwork (F4) was hypothesised.

Workforce involvement in a behaviour-based safety (BBS) programme engenderedtrust in co-workers and management according to DePasquale and Geller (1999). They


also found that accountability was a predictor of involvement in a BBS programme. Forthese reasons, and the possibility that team leaders may make team members accountableby being ‘law enforcers’ (Fleming et al., 1996), a structural path from supervisorinvolvement (F5) to accountability (F8) was specified. It was also hypothesised that therewould be a relationship between propensity to engage in risk-taking behaviour andfeelings of responsibility, so this path was specified in the model. Finally, a feedback loopwas inserted, from responsibility (F6) to senior management commitment (F1). Thisfeedback loop completes a hypothesised cycle of improvement in safety climate. Whenconsidering the model as a whole, the main proposition tested is that senior managers canreduce risk-taking behaviours by investing in resources for knowledge and training. Inturn, this will enhance feelings of responsibility for safety among workforce members andthrough a series of affective responses, will build perceptions of senior managementcommitment. As stated, this proposition is cyclical. The remainder of the present paperdescribes an empirical study which was designed to test this model.

7 Method

7.1 Organisational context and participants

The research context for this study was a UK power-generating company. The participantswere 1023 workforce members from six conventional UK power stations. A relatively flatmanagement structure is in place on those stations, team members (i.e. the workforce)report to a team leader who reports to a functional manager. Above this layer ofmanagement lies the Station Manager. Safety climate questionnaires were administeredto the workforce during working hours and completed on company premises betweenMarch and November 1999. The total response rate was 87%.


Figure 1 Theoretical model of safety climate (model 1). For clarity, the individual items are notdisplayed

7.2 Measures

Workforce safety climate perceptions were collected using the HSE Climate Survey tool(Health and Safety Executive, 1997), which was developed by the UK regulator (Healthand Safety Executive) as a standardised tool for measuring safety climate across all UK industry sectors. It has been used in over 400 companies since commercialisation.However, unlike other measures of safety climate only a superficial evaluation has beenconducted of this tool (Health and Safety Executive, 2002). Although an analysis of thetool’s psychometric properties has been conducted (Yule et al., under review), there arecurrently no published results or models emanating from data collected using the tool. TheHSE Climate Survey tool consists of 71 statements to which responses are made on a fivepoint Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree). Some itemsare negatively worded to balance the questionnaire.

7.3 Analytic strategy

Structural Equation Modelling (SEM) is ideally suited to exploring theoretical paths ofinfluence among safety climate dimensions and testing theoretical models. The techniquehas been used on a number of occasions in the safety sciences (Cheyne et al., 1999; Coyleet al., 1995; Griffin and Neal, 2000; Mearns et al., 2001; Tomas et al., 1999). SEM wasachieved using EQS for windows version 5.5 (Bentler and Wu, 1995).

8 Results

8.1 Exploratory factor analysis

The factorial structure of the safety climate questionnaire was tested with raw responses(n=1023) to preserve statistical power. Responses to negatively worded statements werereversed so higher scores reflected stronger perceptions of safety climate. Exploratoryprincipal components analysis (PCA) with varimax rotation was performed using SPSS9. A criterion of 0.45 was set for individual questionnaire items to significantly load onfactors. The factor analysis identified an eight-factor solution that incorporated 47questionnaire items and accounted for 47% of the variance in safety climate scores. Thiswas deemed to be a statistically acceptable foundation to test for structural linkagesbetween factors using this data set. The eight factors reflect the underlying dimensions ofsafety climate as measured by the HSE tool in this study. In order to interpret and labelthese factors a group of human factors experts and safety practitioners were given therelevant questions grouped as factors and asked to provide a unifying label for each factor.A thematic analysis was then used to generate a set of generally accepted factor labels.The resulting factor labels are presented as Table 1, along with example questions and theproportion of total variance accounted for by each factor.

8.2 Confirmatory factor analysis

In accordance with Byrne (1994), a judicial selection of items from each factor was madeto represent that factor in the modelling. These factors can be conceptualised as subscalesof safety climate and are used as predictor variables in the modelling.


The structural paths detailed above were imposed according to theory. All disturbance anderror variances were freely estimable as were all structural path coefficients betweenlatent variables. The model was tested using EQS for Windows version 5.5 (Bentler andWu, 1995). The solution converged in six iterations with a final function of 1.12507. Thechi-square for this run was 888.8 based on 316 degrees of freedom and with an associatedprobability less than 0.001. The CFI and robust CFI were above the critical value of 0.9, the fit index of choice, CFI being 0.905. See Table 2 for complete information aboutfit indices. The hypothesised structural model of safety climate provides an acceptable fitto the data.

8.3 Respecified structural model

Model 2 uses the hypothesised model in Figure 1 as a basis for respecification. Resultsfrom the Wald test suggested the redundancy of three structural paths: from seniormanagement to risk-taking behaviour (F1 → F2); supervisor involvement to seniormanagement (F5 → F1), and supervisor involvement to accountability (F5 → F8). Thesepaths were subsequently dropped from the analysis in the interests of parsimony as even


Table 1 Safety climate factors and example items

Factor Label Example item % total Cronbach’svariance alpha

1 Senior management Productivity is usually seen as 11.0% 0.91commitment more important than health

and safety

2 Risk-taking Some jobs here are difficult 9.1% 0.85behaviours to do safely

3 Knowledge and People here do not remember 5.9% 0.77training much of the health and safety

training which applies to their job

4 Teamwork I trust my workmates with my 5.6% 0.84health and safety

5 Supervisor

involvement My immediate boss is receptive to 5.2% 0.78ideas on how to improve healthand safety

6 Responsibility I fully understanding the health 4.2% 0.74and safety risks with the work forwhich I am responsible

7 Safety system The Permit to Work system is over 3.2% 0.69the top given the real risks of someof the jobs it is used for

8 Accountability People who cause accidents here 2.4% 0.60are not held sufficientlyaccountable for their actions

though they were theoretically plausible, they did not add to the fit of the model. Themultivariate LaGrange Multiplier test suggested several paths to be added of which sixwere chosen for inclusion on theoretical grounds. The majority of these paths wereexpected on the basis of the factor covariation matrix from the initial exploratory factoranalysis and appeared intuitively plausible.

The model was revised nine times to include the suggested new paths and removethose that did not add to the fit of the model to the data. The solution converged in fiveiterations with a function of 0.71689. The chi-square for this final run was 566.3 basedon 305 degrees of freedom and with an associated probability less than 0.001. All four fit indices for the model were above the acceptable level of 0.9, the fit index of choice,CFI, being 0.957. The respecified model is shown as Figure 2.

A table of fit indices for the two models discussed in this study is shown as Table 2. Thistable facilitates comparison of the goodness of fit of the two models tested. Fit indices arebased on the chi-square value of the model under scrutiny. EQS provides several differentfit indexes of which four are utilised in this study: CFI, NFI, NNFI and robust CFI. Thecomparative fit index (CFI) is considered the most reliable [51] and is the fit index ofchoice for this study. The proportion of total variance accounted for by endogenousvariables was 53%.


Figure 2 Respecified model of safety climate (model 2), CFI = 0.957. Standardised pathcoefficients are given. For clarity, the individual items and associated errors are not displayed

Table 2 Indices of fit for the theoretical and respecified models of safety climate

Model �2 p CFI NFI NNFI Robust CFI

1 (theoretical) 888.8 316df < 0.001 0.905 0.860 0.894 0.905

2 (respecified) 566.3 305df < 0.001 0.957 0.911 0.950 0.957

9 Discussion

Table 2 shows that the fit indices increased from models 1 (theoretical) to 2 (respecified).Model 2 provides a foundation for identifying and exploring the important relationshipsbetween safety climate variables such as the role of management, knowledge andaccountability.

Only three of the a priori specified paths were deemed non-significant and droppedfrom the analysis. This means that the majority of the initial theoretical paths of influencewere imposed correctly. As they were done so in line with literature and previous researchthis is reflective of the validity of the data set and final model. The majority of added pathswere added on an intuitive basis as some of the links have not been documented inprevious studies. With hindsight, several of the added paths could have been included inthe initial theoretical model but were not in the interests of parsimony. The added pathswere all in line with the results of the principle components extraction and reflectrelationships between factors that show high individual correlations.

The respecified model shows knowledge of health and safety as a key mediatorbetween perceptions of senior management commitment and responsibility for safety. Theinfluence of knowledge and training on reducing risk-taking behaviours was one of thestrongest in the model (0.65, the path coefficient is positive because the scale used wasrecoded so that positive perceptions of risk-taking behaviour equated to a lower frequencyof risk-taking behaviours). Knowledge and training also had direct relationships withresponsibility and accountability. The hypothesised direct relationship between seniormanagement commitment and risk-taking behaviour was not supported which providesfurther evidence of that relationship between these variables being mediated. In two pathsadded in the respecification stage, knowledge was also found to have a significant impacton the use of the safety system and on the level of teamwork experienced. It is plausiblethat knowledge may improve team cohesiveness and performance, reduce the number oferrors made by team members, and increase the capacity for the team to capture error.

Evidence for a reciprocal relationship between senior management and supervisorswas not found. Instead, perceptions of supervisor involvement were found to influenceperceptions of the commitment of senior management. This was a strong relationship inthe model at 0.60 and supports the notion that supervisors must enact the safety policy andvision laid down by senior managers. The degree of senior management commitment tosafety perceived by the workforce may vary as a function of how well supervisors achievethis. O’Dea (2002) found the same effect between perceptions of supervisors and OffshoreInstallation Managers (OIMs, i.e. the site manager) in a study conducted in the UK oil and gas sector. If the supervisor is perceived to do this poorly, or has a low level ofinvolvement with the team, then the messages of strong commitment to safety from seniormanagement may not be received by the workforce. In short, relaying the commitment tosafety of senior management to the workforce is one of the key roles of supervisors. Therespecified model also shows that supervisor involvement has an impact on knowledge andtraining, again emphasising the central role of knowledge in the model.

Accountability for safety is significantly influenced by knowledge and training (0.56)and to a lesser extent individual responsibility (0.22). These relationships seem logical, asaccountability cannot be genuinely felt unless workers have both technical knowledge and knowledge of what is and is not acceptable working practice. It is also logical that workers who feel responsible for health and safety will also feel more accountable,


partly because they will show a desire for job characteristics such as enrichment andenlargement that boost their responsibility and make them more accountable (Hackmanand Oldham, 1980).

Perceptions of risk-taking behaviours were found to have an inverse influence onresponsibility (�0.34). This relationship may seem counter-intuitive at first as it suggeststhat fewer risk-taking behaviours are associated with less responsibility. This is not likelyto be the case because the occasions where risk-taking is seen as having high valence are the very occasions where the workforce feels that the procedures are laborious, over the top and time consuming. The workforce also knows that the procedures are driven bylegislation and policy that is there to protect them and as a result they may be dissatisfiedwith the practicalities of some procedures but still follow them (i.e. do not take risks)through a higher source of motivation. The tested model also provides evidence for thehypothesised feedback loop as feelings of responsibility for safety have a positiverelationship with perceptions of management commitment.

10 Conclusions

The present paper outlines the development of a model of safety climate that stipulatedmanagement commitment and involvement as input variables, and risk-taking behaviouras an outcome. The fit of this model to a safety climate database was tested using EQSand then respecified. The respecified model confirmed our a priori assumption that the relationship between senior management commitment and worker risk-takingbehaviour was mediated by knowledge and training. In turn, reducing the level of risk-taking behaviour had a positive impact on workers’ appraisals of senior managementcommitment to safety through the mediating variable of responsibility for health andsafety. The involvement of supervisors was also found to predict perceptions ofknowledge and training.

The model has applications for both practice and research. In practice, the modelstipulates that senior managers can reduce risk-taking behaviours at work in two ways.Firstly, they can invest in the knowledge and training of their workforce so the workforcemay understand the risks inherent in high hazard work. Workers are more likely to followprocedures when faced with the option of taking a risk if they understand why thoseprocedures are in place and have a more complete picture of the risks that they will beexposed to if they do not follow procedures. Secondly, senior managers can encouragesupervisors to be more involved in safety activities. Suggested behaviours for supervisorsinclude being receptive about workforce ideas about ways to improve health and safety,implementing suggestions from the workforce, demonstrating a duty of care andproviding assistance if workers ask for help. Transactional leadership behaviours (Bass,1998) such as those have already been found to be effective for supervisors in reducinginjury rates at work (Kivimaki et al., 1995). Managers should note that not investing intraining will have a negative effect and may actually encourage risk-taking as workerswill not feel that senior managers are committed or interested in the welfare of theiremployees.

For research purposes, the tested model provides a framework which can be used toguide more focused research on the processes of management commitment and supervisorinvolvement on safety. The model establishes ‘risk-taking behaviours’ as a viable


outcome in safety research. One limitation of the current study is that the outcomemeasure is self-report. Self-report measures are not in themselves unreliable but there isa potential to underestimate engagement in undesirable behaviours such as risk-taking.Future research can build on the findings here by establishing independent measures ofrisk taking and incorporating them in the model. One way of achieving this would be toobserve the actual numbers of risks taken in a set period of time and establish whetherworkers who perceived senior managers to be committed and supervisors to be involvedin safety actually took fewer risks than workers who did not think that managers werecommitted and supervisors involved. Despite this caveat, there are strong real worldapplications of the research presented here for managers and supervisors who want toimprove the state of safety in their organisations.

References

Andriessen, J.H.T.H. (1978) ‘Safe behaviour and safety motivation’, Journal of OccupationalAccidents, Vol. 1, pp.363–376.

Bass, B.M. (1998) Transformational Leadership: Industrial, Military, and Educational Impact,Mahwah, NJ: Lawrence Erlbaum Associates.

Bentler, P.M. and Wu, E.J.C. (1995) EQS for Windows User’s Guide, Encino, CA: MultivariateSoftware, Inc.

Brown, K.A., Willis, P.G. and Prussia, G.E. (2000) ‘Predicting safe employee behaviour in thesteel industry: development and test of a socio-technical model’, Journal of OperationsManagement, Vol. 18, No. 4, pp.445–465.

Brown, R.L. and Holmes, H. (1986) ‘The use of a factor analytic procedure for assessing thevalidity of an employee safety climate model’, Accident Analysis and Prevention, Vol. 18, No. 6, pp.445–470.

Byrne, B.M. (1994) Structural Equation Modelling with EQS and EQS/Windows: Basic Concepts,Applications and Programming, Thousand Oaks, CA: Sage.

Cheyne, A., Cox, S., Oliver, A. and Tomas, J.M. (1998) ‘Modelling safety climate in the predictionof levels of safety activity’, Work and Stress, Vol. 12, No. 3, pp.255–271.

Cheyne, A., Tomas, J.M., Cox, S. and Oliver, A. (1999) ‘Modelling employee attitudes to safety:a comparison across sectors’, European Psychologist, Vol. 4, No. 1, pp.1–10.

Clarke, S. (1999) ‘Perceptions of organisational safety: implications for the development of safetyculture’, Journal of Organisational Behaviour, Vol. 20, No. 2, pp.185–198.

Cox, S. and Cox, T. (1991) ‘The structure of employee attitudes to safety: a European example’,Work and Stress, Vol. 5, No. 1, pp.93–106.

Cox, S. and Flin, R. (1998) ‘Safety culture: philosopher’s stone or man of straw?’, Work and Stress,Vol. 12, No. 3, pp.202–216.

Coyle, I., Sleeman, S. and Adams, D. (1995) ‘Safety climate’, Journal of Safety Research, Vol. 22,No. 4, pp.247–254.

Dedobbeleer, N. and Beland, F. (1991) ‘A safety climate measure for construction sites’, Journalof Safety Research, Vol. 22, No. 2, pp.97–103.

Dedobbeleer, N. and Beland, F. (1998) ‘Is risk perception one of the dimensions of safety climate?’in A. Freyer and A. Williamson (Eds), Occupational Injury: Risk Prevention and Intervention,London: Taylor & Francis, pp.73–81.

DePasquale, J.P. and Geller, E.S. (1999) ‘Critical success factors for behaviour-based safety:a study of twenty industry-wide applications’, Journal of Safety Research, Vol. 30, No. 4,pp.237–249.


Diaz, R. and Cabrera, D. (1997) ‘Safety climate and attitude as evaluation measures oforganisational safety’, Accident Analysis and Prevention, Vol. 29, No. 5, pp.643–650.

Donald, I. and Canter, D. (1994) ‘Employee attitudes and safety in the chemical industry’, Journalof Loss Prevention in the Process Industries, Vol. 7, No. 3, pp.203–208.

Fleming, M.T., Flin, R.H., Mearns, K. and Gordon, R.P.E. (1996) The Offshore Supervisor’s Rolein Safety Management: Law Enforcer or Risk Manager. Paper presented at the 3rd InternationalConference on Health, Safety and Environment in Oil and Gas Exploration and Production,New Orleans (SPE paper 35906).

Flin, R., Mearns, K., Fleming, M. and Gordon, R. (1996) Risk Perception and Safety in the OffshoreOil and Gas Industry. (OTH 94454), Suffolk, HSE Books.

Flin, R., Mearns, K., O’Connor, P. and Bryden, R. (2000) ‘Measuring safety climate: identifyingthe common features’, Safety Science, Vol. 34, No. 1, pp.177–192.

Griffin, M.A. and Neal, A. (2000) ‘Perceptions of safety at work: a framework for linking safetyclimate to safety performance, knowledge and motivation’, Journal of Occupational HealthPsychology, Vol. 5, No. 3, pp.347–358.

Guldenmund, F.W. (2000) ‘The nature of safety culture: a review of theory and research’, SafetyScience, Vol. 34, No. 1, pp.215–257.

Hackman, J.R. and Oldham, G.R. (1980) Work Redesign, London: Addison Wesley.

Health and Safety Executive. (1997) The Health and Safety Climate Survey Tool: HSE.

Health and Safety Executive. (2002) Evaluating the Effectiveness of the Health and SafetyExecutive’s Health and Safety Climate Tool (No. 042). Norwich: Health and Safety Executive.

Hofmann, D.A. and Stetzer, A. (1996) ‘A cross-level investigation of factors influencing unsafebehaviours and accidents’, Personnel Psychology, Vol. 49, No. 2, pp.307–339.

Kivimaki, K., Kalimo, R. and Salminen, S. (1995) ‘Perceived nuclear risk, organisationalcommitment, and appraisals of management: a study of nuclear power plant personnel’, Risk Analysis, Vol. 15, No. 3, pp.391–396.

Kozlowski, S.W.J. and Doherty, M.L. (1989) ‘Integration of climate and leadership: examinationof a neglected issue’, Journal of Applied Psychology, Vol. 74, No. 4, pp.443–449.

Lee, T. (1998) ‘Assessment of safety culture at a nuclear reprocessing plant’, Work and Stress, Vol. 12, No. 3, pp.217–237.

Marek, J., Tangenes, B. and Hellesoy, O.H. (1985) ‘Experience of risk and safety’, in O.H. Hellesoy(Ed.), Work Environment Statfjord. Work Environment, Health and Safety on a North SeaPlatform, Oslo: Universitetsforlaget.

Mearns, K., Flin, R., Fleming, M. and Gordon, R. (1997) Human and organisational factors inoffshore safety (OTH 543), Suffolk, Offshore Safety Division, HSE Books.

Mearns, K., Flin, R., Gordon, R. and Fleming, M. (1998) ‘Measuring safety climate on offshoreinstallations’, Work and Stress, Vol. 12, No. 3, pp.238–254.

Mearns, K., Whitaker, S.M. and Flin, R. (2001) ‘Benchmarking safety climate in hazardousenvironments: a longitudinal, interorganisational approach’, Risk Analysis, Vol. 21, No. 4,pp.771–786.

Niskanen, T. (1994) ‘Safety climate in the road administration’, Safety Science, Vol. 17, No. 4,pp.237–255.

O’Dea, A. (2002) Site Managers and Safety Leadership in the Offshore Oil and Gas Industry,University of Aberdeen.

Rundmo, T. (1994) ‘Associations between safety and contingency measures and occupationalaccidents on offshore petroleum platforms’, Scandinavian Journal of Work and EnvironmentalHealth, Vol. 20, pp.128–131.

Rundmo, T. and Hale, A.R. (In press) ‘Managers’ attitudes towards safety and accident prevention’,Safety Science.


Rundmo, T. and Sjöberg, L. (1998) ‘Risk perception by offshore oil personnel during bad weatherconditions’, Risk Analysis, Vol. 18, No. 1, pp.111–118.

Safety culture. (No. No. 75-INSAG-4)(1991), Vienna: International Atomic Energy Agency.

Simard, M. and Marchand, A. (1994) ‘The behaviour of first-line supervisors in accident preventionand effectiveness in occupational safety’, Safety Science, Vol. 17, No. 3, pp.169–185.

Simard, M. and Marchand, A. (1995) ‘A multilevel analysis of organisational factors related to thetaking of safety initiatives by work groups’, Safety Science, Vol. 21, No. 2, pp.113–129.

Smith, M., Cohen, H., Cohen, A. and Cleveland, R. (1978) ‘Characteristics of successful safetyprograms’, Journal of Safety Research, Vol. 10, No. 1, pp.5–15.

Thompson, R.C., Hilton, T.F. and Witt, L.A. (1998) ‘Where the safety rubber meets the shop floor:a confirmatory model of management influence on workplace safety’, Journal of SafetyResearch, Vol. 29, No. 1, pp.15–24.

Tomas, J.M., Melia, J.L. and Oliver, A.M. (1999) ‘A cross validation of a structural equation modelof accidents: organisational and psychological variables as predictors of work safety’, Workand Stress, Vol. 13, No. 1, pp.49–58.

Williamson, A., Feyer, A., Cairns, D. and Biancotti, D. (1997) ‘The development of a measure ofsafety climate: the role of safety perceptions and attitudes’, Safety Science, Vol. 25, No. 1,pp.15–27.

Witt, L.A., Hellman, C. and Hilton, T.F. (1994). Management Influences on Perceived Safety, Paperpresented at the American Psychological Society Annual Meeting, San Francisco, CA.

Yukl, G. (1998) Leadership in Organizations, (4th ed.), NJ, Prentice Hall Inc.

Yule, S. (2003) Senior Management Influence on Safety in the UK and US Energy Sectors,University of Aberdeen.

Yule, S.J., O’Connor, P. and Flin, R. (Under review) ‘Testing the structure of a generic safetyclimate survey instrument: Implications for applied research and practice’.

Zohar, D. (1980) ‘Safety climate in industrial organisations: theoretical and applied implications’,Journal of Applied Psychology, Vol. 65, No. 1, pp.96–102.

Zohar, D. (2000) ‘A group-level model of safety climate: testing the effect of group climate onmicroaccidents in manufacturing jobs’, Journal of Applied Psychology, Vol. 85, No. 4,pp.587–596.

Zohar, D. (2002) ‘Modifying supervisory practices to improve sub-unit safety: a leadership-basedintervention model’, Journal of Applied Psychology, Vol. 87, No. 1, pp.156–163.

Notes

1 This manuscript is based on a paper previously presented at the annual SIOP Conference, San Diego, 2001.

2 In the PTW system, an authorised person issues a printed ‘permit’ to workers once riskassessments for the job have been completed. Work can only commence once the permit has beenissued.


The Role of Management and Safety Climate in Preventing Risk Taking at Work

Documents

Transcript of The Role of Management and Safety Climate in Preventing Risk Taking at Work